Close #4206.

This should correct error handling cases.

This is not used anymore.

This was meant for interleaving RTP.

In append mode, all write operations must occur at the end of the file.
Nevertheless the initial read offset is the beginning of the file.

mingw-w64 FORCEINLINE definition includes 'extern'

We have to count for "Bold", "Normal" and "Italic" ending in registry

Close #4949

Probing is buggy and slow (and it runs for every input item!).

 * fix memory leak of image attribute in case of error,
 * fix clobber of video format structure in case of error,
 * print unknown XVideo format error only once per adaptor
   (rather than once per VLC format per adaptor)

var_LocationParse() enforces change_safe().

Just about every other capture mechanism allows selecting the device
from the MRL, so we might as well support it for DVB. The playlist will
only work on a specific system, but it should still be safe.

Signed-off-by: Felix Paul Kühne <fkuehne@videolan.org>

And fix the crash of http://streams.videolan.org/issues/mkv/invalid_EBML_number_size_tag.mkvSigned-off-by: Jean-Baptiste Kempf <jb@videolan.org>

It might be better (dunno if it would be simpler) to have a way to
specify some kind of profile.

Since this is the default AR for Home Theaters Projectors

When we haven't found the right font, use arial.ttf

Close #4946

Close #5495

because x-axis was confusing to some users, notably with the "Magic
mouse"

Close #5421

Mixed endianess is only an issue if the component values are not
aligned inside octets. This fixes Debian #604687.

(VLC simply lost one sample - no memory errors.)

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

Signed-off-by: Rémi Denis-Courmont <remi@remlab.net>

That was conflicting with the one in vlc_network.h

Instead of asking the user for where to place it...
I think %tmp% should always be writtable.

Width and height are independent for step-wise frame sizes (yes,
enumerating is slow). Continuous frame sizes are step-size with one
pixel steps for both dimensions (yes, enumerationg is very slow).

A few device drivers use either of those two types, though the discrete
type is much more common.